Using a genotyping-by-sequencing (GBS) approach to elucidate population structure in Garry Oak (Quercus garryana)

A pilot study was performed to assess the usability of the genotyping-by-sequencing (GBS) method to perform population genetic analyses across the species range of Garry oak (Quercus garryana). Ninety-two individuals representing 11 populations were sequenced in a single multiplexed GBS library, which resulted in 18.1 Gigabases of sequence data covering 20 Megabases of unique sequences for 87 of the 92 individuals sequenced, which is estimated to represent ~2% of the genome. 20,751 high-quality single nucleotide polymorphisms (SNPs) were obtained, which were used to determine population differentiation and genetic structure. A more-refined set of 2,194 SNPs was developed and used for Bayesian genetic clustering analysis. Clustering analyses revealed two distinct genetic clusters, one corresponding to Q. garryana var. garryana and the other to both vars. semota and breweri. Some evidence supports the presence of additional hierarchical structure within both of these clusters, although there was no evidence to support the taxonomic division between vars. semota and breweri. Overall FST averaged 0.091, and was strongly correlated with physical distance (R²=0.93, p<0.0001). Pair-wise FST was significantly higher between populations of different genetic clusters than among them. Heterozygosity was negatively correlated with latitude (R²=0.72, p=0.001), and was higher in vars. semota and breweri than in var. garryana (p<0.0001). Inbreeding coefficient was also negatively correlated with latitude (R²=0.65, p=0.003). Genetic distance was positively correlated at distances up to 400 km (Mantel test, 10,000 permutations; p<0.001) and negatively correlated beyond 1000km (p